138 replies to this topic

[Nightbird]

https://mwomercs.com...onstant-losses/

Losing too much? Win more with these suggestions. Winning too much? Lose more with these suggestions. (Yes, you can win so hard and often that you fall asleep)

[Nightbird]

https://mwomercs.com...o-small-groups/

Suggested MM will weigh groups as more than the sum of their parts, allowing solos, small groups, and big groups to be in one match.

[Nightbird]

I updated the OP to be simpler and more concise.


https://mwomercs.com...y-morning-ques/

The requested changes include dynamic teams, which can not only put odd sided teams together, but also launch with less than 24 players. If your off-peak population queue only allows a 8vs9 match, the system will launch your game.

[LordNothing]

the reason the matchmaker is so broke is because it depends on the broken psr system. the bubbleup effect needs to disapear around t2 and be negative at t1. then perhaps the matchmaker can ensure that the players are evenly matched and still insulate new players from the shark pit.

[Nightbird]

LordNothing, on 14 June 2018 - 06:28 PM, said:

the reason the matchmaker is so broke is because it depends on the broken psr system. the bubbleup effect needs to disapear around t2 and be negative at t1. then perhaps the matchmaker can ensure that the players are evenly matched and still insulate new players from the shark pit.

While this would improve things, it wouldn't quite give you fair matches because
1. Tiers are still buckets, the skill gap between the bottom of T1 to the top of T1 will be as wide as T5-T2.
2. People may want to skill up mech, use non-serious mechs, if all these people are by chance put on one team > stomp
3. There are long, mid, short range loadouts. MM by chance puts all short range mechs on one side and long range on the other, on Polar or Mining, and stomp

Edited by Nightbird, 15 June 2018 - 05:52 AM.

[Haipyng]

I really like the sound of it, especially for GP. I don't believe the MM there calculates anything other than tonnage and group sizes when making up teams. The trouble is populations are so small in GP. Would that affect it? Do we have a hope that PGI can wrangle a complicated MM system given the trouble it has had with QP MM, which by comparison to GP MM is totally fair and balanced.

[MischiefSC]

The problem with dynamic teams is it's a harder set of factors to balance than we already have, which is already too much to do accurately with any of the current systems.

Team size differences, even by 2 mechs, is enough to make a dramatic swing. While better players can carry hard enough to make up for the difference it would require maximum tryhard behaviors every match. This also falls apart the moment you get down the curve from 'comp/near comp level player skill'.

It's possible but it seems like it would be way more of a mess than just trying to balance 12v12, which is already outside the scope of anything PGI has done before.

Just start with an accurate rating for each player (something akin to Elo) separate for group and solo queues and then get some sort of modifier for mechs and loadouts. That alone would let you play far more fast and loose with tonnage and increaes accuracy by a mile - if you've got the population depth to leverage it.

There's a different level of commitment required from the player to win in a 'fewer team members' situation. It can happen but it's less predictable. You're wanting the matchmaker to be able to dynamically place players based on their challenge tier of performance and not average tier. Yeah, large enough sample sizes can account for a lot but the individual matches are going to swing absolutely wild and I don't think it would give a better match by match experience than we have now.

[Nightbird]

I agree that it would be hard, but it can and must be done. When a team like KCOM which you are part of is running a 8-12 man, there is no team the MM can put together in queue to create a fair match. Do we just say, oh, KCOM gets to club seals, 100% loss chance matchups are ok, or do we hold your team until another skill group arrives, be it 5 minutes, hours, or days...?

If fair matches are important, and you also want reasonable wait time, there are 0 other solutions.

[Nightbird]

https://mwomercs.com...g-fun-is-in-qp/

A challenging match where both teams have the same shot at winning.

[MischiefSC]

Nightbird, on 15 June 2018 - 11:47 AM, said:

I agree that it would be hard, but it can and must be done. When a team like KCOM which you are part of is running a 8-12 man, there is no team the MM can put together in queue to create a fair match. Do we just say, oh, KCOM gets to club seals, 100% loss chance matchups are ok, or do we hold your team until another skill group arrives, be it 5 minutes, hours, or days...?

If fair matches are important, and you also want reasonable wait time, there are 0 other solutions.

Depends on the environment. In theory FW should have 0 matchmaker because it's factions fighting - you want to take a world you have to beat the best they have to offer.

If it's group queue you've got tonnage limits and playing in a 12man really sucks. You're all in lights and mediums and you have to be full try hard mode to have a chance at a win. It certainly stacks odds against the 12man.

As a matchmaker option, well regardless of what you do with the matchmaker it can only build a match based on who is in queue. Punishing successful teams or successful players for being successful just drives player churn. Ideally you want enough people in queue to skill match. Without that if doesn't really matter.

If your logic is that a 12man of KCom in the group queue (which you'd never see, we would be in FW) is that we would play against a mixed group of lower average Elo with potentially over 12 players then you're trying to effectively predict how the >12 team would play vs our 12man. This goes beyond just player and mech performance variables but I to how well that team leverages additional players. You're adding in a *potential* magnifier for teamwork. It's not just a straight HP/DPS buff to one side.

That would be messy as ****. You're better off having something like dynamic tonnage limits based on Elo, so a 4man of bads get more tonnage than a 4man of Emp.

Maybe I just need to play around with the math a bit but I'm struggling to see where you could make a good dynamic prediction model. You could easily make a very bad one - and maybe that's still an okay solution. We already have a bad system in place, a bad one that feels more dynamic might still be better, I'm just not sure what it would look like.

[Nightbird]

MischiefSC, on 16 June 2018 - 11:31 AM, said:

1) Punishing successful teams or successful players for being successful just drives player churn. Ideally you want enough people in queue to skill match. Without that if doesn't really matter.

2) Maybe I just need to play around with the math a bit but I'm struggling to see where you could make a good dynamic prediction model. You could easily make a very bad one - and maybe that's still an okay solution. We already have a bad system in place, a bad one that feels more dynamic might still be better, I'm just not sure what it would look like.

I'll focus on just two point in your post if that's ok.

1) I've played on skilled big teams in both QP and FP (as part of NS, 42, EVIL), I'm not trying to single KCOM out. When we face another good team, I personally feel my heart pumping, I'm excited, I'm happy when we win and when we lose. I feel that a lot of skilled groups have stopped playing because 99% of the time, it is beating a team where I have literally fallen asleep with my mech walking into a wall and we still win. I don't see it as punishing a skilled team as much as saving it from that eventual cliff where players after becoming a good pilot also become disinterested due to the lack of challenge. Feel free to disagree with me here, all my personal views.

2) From an intuitive point of view, I'm sure when you see a team on the other side you can instantly predict that the final score is going to be (12 to 0,1,2) or (48 to 15-20). Add two players to the other side and maybe it'll be (14 to 2-6) or (56 to 20-36). Add four players and maybe it'll be (18 to 6-12) or (64 to 30-48). The mathematical model's goal is to change this from intuition to a computer programmable algorithm, with the goal of ensuring the stronger team has the possibility of losing. I would probably agree that if the smaller team is put into a position where it is rolled badly, the MM failed. The force magnifier that comes with having more team members can definitely be predicted though. The team values would be calculated by the sum of their parts, but the larger team will need a multiplier based on the difference in team numbers.

Edited by Nightbird, 16 June 2018 - 12:03 PM.

[MischiefSC]

Nightbird, on 16 June 2018 - 12:01 PM, said:

I'll focus on just two point in your post if that's ok.

1) I've played on skilled big teams in both QP and FP (as part of NS, 42, EVIL), I'm not trying to single KCOM out. When we face another good team, I personally feel my heart pumping, I'm excited, I'm happy when we win and when we lose. I feel that a lot of skilled groups have stopped playing because 99% of the time, it is beating a team where I have literally fallen asleep with my mech walking into a wall and we still win. I don't see it as punishing a skilled team as much as saving it from that eventual cliff where players after becoming a good pilot also become disinterested due to the lack of challenge. Feel free to disagree with me here, all my personal views.

2) From an intuitive point of view, I'm sure when you see a team on the other side you can instantly predict that the final score is going to be (12 to 0,1,2) or (48 to 15-20). Add two players to the other side and maybe it'll be (14 to 2-6) or (56 to 20-36). Add four players and maybe it'll be (18 to 6-12) or (64 to 30-48). The mathematical model's goal is to change this from intuition to a computer programmable algorithm, with the goal of ensuring the stronger team has the possibility of losing. I would probably agree that if the smaller team is put into a position where it is rolled badly, the MM failed. The force magnifier that comes with having more team members can definitely be predicted though. The team values would be calculated by the sum of their parts, but the larger team will need a multiplier based on the difference in team numbers.

I rarely play anymore for the same reasons. I get what you're saying, I do.

However, it's critical to understand that you're a freak. I mean that in a good way - what you want and enjoy is atypical. It's why you're in the handful percentile. What you enjoy drives you to be challenged and improve. You enjoy not playing the game so much as playing the game *better*.

That's about 12% of the population. Little more, little less depending on environment. Most of those enjoy that sometimes but not all the time. Appetite for risk, appetite for challenge, validation through self-improvement, etc. It's a lot of psychology ****.

You're talking about a MM designed to functionally predict a players psychological profile - is this someone who wants challenged, is this someone who is unwilling/unable to adapt and needs carried, etc. Your performance and appetite for risk isn't static. It changes through out the day. Yeah, the group v group matches when everyone is on point are amazing. It's why I'm still here at all. However sometimes I just want to see how the Banshee S3 works in QP stock. I'm not always on point. Sometimes my feet hurt from standing to play with my Vive Pro all day, I just want to sit on my *** and shoot stompy robbits for a bit.

I like the concept. The problem is that you're trying to balance a MM on optimal performance both solo and in grouo, not average performance.

However, that doesn't mean it can't be made to work. It's a fascinating concept - a MM that adjusts for who's in the queue. First you would need opt-in for "I'm okay with dynamic MM if it reduces wait times even if it puts me in a more challenging environment". Then you have dynamic rewards - being on the big team pays the same, being in the small team pays more. If you don't reward opting in very few will.

It's going to be cludgy AF and will take a huge sample size to even start to get accurate, you'd need an excellent Elo system and you would absolutely need something to value mechs and builds - your Deathstrike carries different from your stock Spider. If it's balanced towards player optimal not player average then tolerances are smaller.

I don't object on principle. It has the potential to be a lot of fun and make the better players feel special and consistently challenged while making the sub par players feel like they ha e the advantage the keep saying they deserve to be "fair".

Just getting an eye twitch trying to figure out how to make the math behind balancing work. How do you even begin to accurately value relative focus fire potential, you've also got different sorts of bads. Someone who's bad at positioning would impact this different than someone who's bad at mechlab. I'd just want to wing it for a while and collect a **** ton of data and start iterating.

[Nightbird]

I read your entire post, I think I understand your viewpoint and you understand mine. I did want to address one point:

MischiefSC, on 16 June 2018 - 01:40 PM, said:

That's about 12% of the population. Little more, little less depending on environment. Most of those enjoy that sometimes but not all the time. Appetite for risk, appetite for challenge, validation through self-improvement, etc. It's a lot of psychology ****.

What portion of the population would you say is willing to learn in the extremely adverse environment we have today? The lesser skilled teams get stomped over and over again, and many people stop trying. We tell them, git gud, accept the challenges, self-improve, etc instead of giving them a handicap as they learn. That is what we have today, is it optimal?

[MischiefSC]

Nightbird, on 16 June 2018 - 02:51 PM, said:

I read your entire post, I think I understand your viewpoint and you understand mine. I did want to address one point:



What portion of the population would you say is willing to learn in the extremely adverse environment we have today? The lesser skilled teams get stomped over and over again, and many people stop trying. We tell them, git gud, accept the challenges, self-improve, etc instead of giving them a handicap as they learn. That is what we have today, is it optimal?

They don't learn. I spend most days tracking trends and behaviors for almost 20k employees across all lines of business from retail sales to phone support and reps to developers, engineers, field service techs, even most tiers of leadership. I work out what works and what doesn't, incentive and bonus tiers, figure out who's cheating and how, etc.

About 12% of people have a Git Gud approach to things. They figure out what works and apply it, adjust and adapt, etc. 12% is actually generous AF. In actual practice it's more link single digits in the population at large but let's assume that the games population is as motivated as the people in my companies strict hiring guidelines.

A handicap doesn't help. It's an outlook - it's not something you can compel. The best way to move someone toward a git gud approach is exposure to someone who's already good and willing to mentor. That still only gets you a tiny number of people to buy in.

The vast majority say the system is broken or have excuses why they can't change or improve and everyone doing better than them is either cheating or a sellout in some way.

Your MM suggestion won't change that or those players. They'll say having and advantage is "fair" and still rage when they lose and not do anything to improve.

[Nightbird]

A game is not a job. Where the employee is at fault if they do not fulfill their responsibilities, the game developer is at fault if they do not create an environment favorable to their players. It's impossible to cater to everyone, but like you said, only a small portion of the population can succeed (have fun) in MWO's current environment. How about opening it up a little. It won't be for everyone all at once, but if today 12% can be serviced by MWO, why not target 20% or 30% with the next iteration.

Edited by Nightbird, 16 June 2018 - 05:54 PM.

[MischiefSC]

Nightbird, on 16 June 2018 - 05:53 PM, said:

A game is not a job. Where the employee is at fault if they do not fulfill their responsibilities, the game developer is at fault if they do not create an environment favorable to their players. It's impossible to cater to everyone, but like you said, only a small portion of the population can succeed (have fun) in MWO's current environment. How about opening it up a little. It won't be for everyone all at once, but if today 12% can be serviced by MWO, why not target 20% or 30% with the next iteration.

Games, jobs, relationships, it's all the same. When confronted by a challenge or losing a contest does someone re-evaluate, evolve, improve and adapt or do they justify, cast blame and drill down on failing behaviors wanting to make them work anyway.

My take is this -

Your idea can serve both. The reality is that the 12% is going to be the group populating the smaller teams and they'll enjoy the challenge and the glory and sense of accomplishment that comes with a clearly defined "hardmode". The bigger team sizes will be driven by the worst parts of the 88%. Most people fall along the spectrum somewhere and most will happily do whatever gets them matches faster and at least a better sense of balance.

It has the potential to drive better habits too. If you drop on a 10man team vs a 12man team and you see some top tier player on your team (probably why you're 2 players down) you're prone to want to coordinate more. It also potentially makes every match feel "unique" instead of expendable.

I like the idea, it has a lot of potential. You need to make it opt-in. Most people will opt in eventually but it needs to be a choice to participate. It's also going to be wonky for a while but a month or so of data collection and on the fly adjustments could make it *feel* more balanced. Doesn't matter if it is, it just has the potential to feel that way. Getting stomped in a 12v12 when clearly the other team was better sucks. Losing in a 12man to an 8man may technically be just as unbalanced but at least you feel like you had an advantage.

Just not sure mechanically the game would support it.

[Nightbird]

MischiefSC, on 17 June 2018 - 06:32 AM, said:

Just not sure mechanically the game would support it.

PGI... do you accept the challenge?

[Nightbird]

https://mwomercs.com...-a-bunch-of-t5/

Nope, just the MM not caring about unskilled un-upgraded mechs as usuall.

[Astrocanis]

MischiefSC, on 17 June 2018 - 06:32 AM, said:

Games, jobs, relationships, it's all the same. When confronted by a challenge or losing a contest does someone re-evaluate, evolve, improve and adapt or do they justify, cast blame and drill down on failing behaviors wanting to make them work anyway.

My take is this -

Your idea can serve both. The reality is that the 12% is going to be the group populating the smaller teams and they'll enjoy the challenge and the glory and sense of accomplishment that comes with a clearly defined "hardmode". The bigger team sizes will be driven by the worst parts of the 88%. Most people fall along the spectrum somewhere and most will happily do whatever gets them matches faster and at least a better sense of balance.

It has the potential to drive better habits too. If you drop on a 10man team vs a 12man team and you see some top tier player on your team (probably why you're 2 players down) you're prone to want to coordinate more. It also potentially makes every match feel "unique" instead of expendable.

I like the idea, it has a lot of potential. You need to make it opt-in. Most people will opt in eventually but it needs to be a choice to participate. It's also going to be wonky for a while but a month or so of data collection and on the fly adjustments could make it *feel* more balanced. Doesn't matter if it is, it just has the potential to feel that way. Getting stomped in a 12v12 when clearly the other team was better sucks. Losing in a 12man to an 8man may technically be just as unbalanced but at least you feel like you had an advantage.

Just not sure mechanically the game would support it.

Psycho babble. People who lose constantly in a game, particularly one that's free, quit. They don't "rise to the challenge", which in this game is hokum because stock mechs suck and even with the tutorial (which, while pretty, isn't much help). When you play 12 games and get stomped 12 times, it's demoralizing.

I work hard and have for many years. My career blossomed and I have been very successful. My friends are from that same type A group. And NOT ONE OF THEM have time to deal with the colossally bad new player experience this game provides. Period. Git gud? That's a trite saying that means "I have a knack for this and screw you." Well, my friends did say "Screw it" with the game. Try telling a bunch of CEO/CTO/CFO to "git gud" at a frigging game some time.

My response to this nonsense is actually me being kind. My friends' responses have been more of a string of four letter words.

In short, most of us don't have the time or the inclination to "git gud" at a computer game, especially one where all of the social aspects of the game exist OUTSIDE the game. At least when I get frustrated with golf, I can have a beer with my friends and get over it. And I don't feel any special requirement to "git gud" at golf either.

[MischiefSC]

Astrocanis, on 18 June 2018 - 02:02 PM, said:

Psycho babble. People who lose constantly in a game, particularly one that's free, quit. They don't "rise to the challenge", which in this game is hokum because stock mechs suck and even with the tutorial (which, while pretty, isn't much help). When you play 12 games and get stomped 12 times, it's demoralizing.

I work hard and have for many years. My career blossomed and I have been very successful. My friends are from that same type A group. And NOT ONE OF THEM have time to deal with the colossally bad new player experience this game provides. Period. Git gud? That's a trite saying that means "I have a knack for this and screw you." Well, my friends did say "Screw it" with the game. Try telling a bunch of CEO/CTO/CFO to "git gud" at a frigging game some time.

My response to this nonsense is actually me being kind. My friends' responses have been more of a string of four letter words.

In short, most of us don't have the time or the inclination to "git gud" at a computer game, especially one where all of the social aspects of the game exist OUTSIDE the game. At least when I get frustrated with golf, I can have a beer with my friends and get over it. And I don't feel any special requirement to "git gud" at golf either.

So you prioritize where you put your effort, like everyone else. What's worth spending time on to get results. Same as everyone else. However in the context of the matchmaker the difference between someone who is theoretically capable of learning to be good at MWO but is unwilling to expend the effort is indistinguishable from someone who is completely incapable of it.

However the willingness to recognize that is what makes the difference. There is no 'knack' for anything other than 'did your initial inclination on how to do something match with what actually works', baring physiological/biological factors like health and fitness impacting your 'knack' for health and fitness related skills.

Getting good at MWO is significantly less difficult than getting good at golf by the way.

However you're mistaking your anecdotal experiences and confirmation bias for how people work. Given the many tens of thousands who play the game and many thousands, over the years tens of thousands who've 'got gud' didn't quit because they had a string of losses (and everyone who's played any PvP game have had strings of losses) your assessment of what happens is also wrong.

Back around to the point though as I said, the difference between someone unwilling and someone unable to put the effort into 'git gud' in MWO is irrelevant to the matchmaker and their behaviors and responses are functionally identical.